{"id":"W4385572319","doi":"10.18653/v1/2023.acl-industry.55","title":"Weighted Contrastive Learning With False Negative Control to Help Long-tailed Product Classification","year":2023,"lang":"en","type":"article","venue":"","topic":"Text and Document Classification Technologies","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"","keywords":"Categorical variable; Categorization; Computer science; Regularization (linguistics); Artificial intelligence; Task (project management); Taxonomy (biology); Machine learning; Pattern recognition (psychology); Natural language processing; Engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003103761,0.0001923192,0.0002201992,0.0003528599,0.0002304618,0.0002513902,0.0006865713,0.00006098707,0.00004002684],"category_scores_gemma":[0.0003437384,0.0001407071,0.00003670181,0.0019705,0.00009482726,0.0005900107,0.0001239362,0.0002025025,0.001132693],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007472343,"about_ca_system_score_gemma":0.00007953156,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001502077,"about_ca_topic_score_gemma":0.0000263096,"domain_scores_codex":[0.9982561,0.00009781831,0.0002519886,0.000666515,0.0003524077,0.0003751216],"domain_scores_gemma":[0.9986219,0.0002700792,0.0001529043,0.0005477857,0.000305717,0.0001015905],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0003267805,0.0002539083,0.09605554,0.00003437322,0.0002605299,0.00004853839,0.00316673,0.0008836312,0.03291015,0.4019188,0.0120097,0.4521313],"study_design_scores_gemma":[0.004155512,0.001316048,0.7373661,0.00009578111,0.00004343254,0.00001321336,0.004754453,0.1572516,0.0761762,0.007968021,0.009643347,0.001216308],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08969402,0.00002126224,0.8821945,0.02057353,0.000124884,0.001052035,0.000002421287,0.004140874,0.002196442],"genre_scores_gemma":[0.9852179,0.000009884465,0.008021231,0.000225358,0.00003350879,0.0002952149,0.000009338469,0.00001347583,0.006174042],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8955239,"threshold_uncertainty_score":0.9996451,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01927232653048668,"score_gpt":0.2567327726888022,"score_spread":0.2374604461583155,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}